Power P is product by voltage V and current I. Checking the measurement error,
P = VI

Differentiating，
dP = VdI + IdV
dP/P = dI/I + dV/V

Therefore，measurement error of power is sum of current and voltage.

I always use DSO for DMM，DC accuracy of DSO is 3%. it is not good. The DSO measurement has repeatability, but cannot decrease absolute errors. I have 3 DMMs of PC510, SD-420C and P-16. They lack one to measure power. I connected an IC clip to change input and output voltage, as I learned science in junior high school.

Input power error PC510 + SD-420C = 1.5%
Output power error P-16 + SD-420C = 3.3%

DMM P-16 restricts the measurement accuracy.

My DMMs accuracy
PC510 0.2% for current
P-16 2.0% for current
SD-420C 1.3% for voltage

Voltage regulator from 0V

I have no power supply available to adjust output near 0V. I had worked a small circuit to control below 1V 1A. So I have tried closed loop control to decrease power dissipation and failed. I rebuilt adopting PNP transistor that I have had 2SA1056 and 2SA562 for NPN. The new TTA004B maximum collector current is 1.5A, while 2SD553 is 7A. But TTAA004B is cheap. TTA004B h_FE breaks down around 0.7A. I decided not to work over 1A. It inflows 0.948A (PC510) @ 0.403V (SD240) in case of load 0.3Ω. The collector loss is,
Pc = 0.951 x 3.94 = 3.75W

Thermal resistor calculus of heat sink

I have heat sinks of thermal resistor 12.5 °C/W. I estimated the thermal resistor of the heat sink and TTA004B is 19.5 °C/W. While I measured the rising temperature by Tanita TT-533 thermometer. The thermal resitor is 8.4℃/W only. The actual thermal resistor might be the intermediate value. It is difficult to measure temperature of surface body.

Electrolytic vs Ceramic capacitor

I measured input/output voltage and output current shown in Photo1. The DC/DC converter is on the blue board. You can see the blue 47uF capacitor on it.

Figure2 and Figure3 show efficiency charts of electrolytic and ceramic capacitor. Fixed 3.3k, 330 and 33Ω resistor inflows 1, 10 and 100mA each approximately.

Ceramic: 200uA output

1) 100mA
Electrolytic is better except 0.8V.
2) 10, 1mA
Ceramic is better over 2.0V. It is impossible to measure inflow current because the value fluctuates.

The actual output voltage drop is shown in Figure4. The ceramic capacitor cannot control flat at 3V3 as well as electrolytic does. But ceramic drops little than electrolytic below 1.4V. The 100mA of Ceramic exceeds 3V3 over 2.4V input.

When input is 2.86V, output fluctuates 214 to 218 uA @ 1.5k load. The ripple is 240mV and the period is 2s. Lowering input 1.027V, output is stable at 212.6 to 212.8 uA. But input current is not continuous but discrete. The pulse period varies for input voltage as shown in Table1. I measured the input pulse current by my built peak current detector that I reported in April 2017 as shown in Photo2. The orange object is ceramic 10uF capacitor on the blue board.

Table1 Input pulse current

Vin [V]	Period [ms]	Iout [uA]
1.028	50	212.5
0.904	134	212.5
0.802*	258	212.4 to 213.4

*) Output voltage is 3.29V.

Conclusion

The ceramic capacitor 3V3 DC/DC converter cannot replace unused two series batteries, because the output voltage exceeds the maximum voltage of TWE. Single battery is preferable.

In case of 1mA and 10mA, input current is not continuous more than input voltage 2.4V. I do not know how input pulse current has a good or bad effect on battery. XBees also have been dissipating discrete pulse current while sleeping. It will not be anxious about pulse current, will it?

Table2, Table3 and Table4 show the values that PC510 and P-16 measured the same current. At a glance, it looks the values correct. But the third digit of them are uncertain. It is necessary to use more precise meters to argue 1% efficiency up/down, I think.

Ref.

Мультимер Sanwa SD420C
PC510

Table2 DMM current digit columns

PC510 [mA]	P-16 [mA]	Deviation [mA]
53.1	52.46	-0.64
25.05	24.83	-0.22
11.40	11.41	0.01
4.86	4.82	-0.04

Table3 DMM current digit columns

PC510 [mA]	P-16 [mA]	Deviation [mA]
6.35	6.3	-0.05
9.62	9.55	-0.07
67.3	66.7	-0.6
99.3	98.1	-1.2
101.4	100.4	-1

Table4 DMM current digit columns

PC510 [uA]	P-16 [uA]	Deviation [uA]
983	989	5
637	641	4
105.7	105.5	-0.6
99.3	98.1	-1.2
101.4	100.4	-0.2